Vocal analysis of emotions during the recall of Self-Defining Memories (SDM) among antisocial population

Little is known about the emergence of emotions through voice among antisocial population (AS). However, vocal cues such as fundamental frequency (F0) are specific indications related to emotional regulation. This ability has been highlighted as deficient among AS population. Therefore, this study aims to analyze the vocal cues of emotions during the recall of five Self-Defining Memory (SDM). We collected 86 SDM from antisocial participants in a forensic medium-security hospital in Belgium. Seven acoustic parameters were analyzed and linked with emotional regulation and empathy abilities. The results will be discussed in the light of the literature.

Summary (500 words)
Background
Social interactions rely, among others, on vocal cues such as speech prosody (Bänziger & Scherer, 2005). Emotional vocal markers influence the expressions of emotions (Scherer, 2003). Despite the increasing number of studies on the features of vocal markers, few have focused on emotions expressed in real-life context (Fontaine, Scherer, & Soriano, 2013). Emotions, and particularly strong ones, build the daily-life experience of individuals through memory, and specifically, Self-Defining Memory (SDM; Martinelli & Piolino, 2009). SDM are memories enclosed in the Autobiographical Memory (AM) and have several roles. Among these, it supports the construction of Self and, through its influence on the self-identity process, has a social dimension (Martinelli & Piolino, 2009; Wagener et al., 2015). Authors have identified dysregulation of emotions as a key-factor for the etiology of aggressive behavior (Gillespie & Beech, 2018; Stinson & Becker, 2013). Therefore, this study aims to investigate the acoustic parameters of vocal cues from naturalistic expressed emotions during the recall of SDM among antisocial forensic patients. Moreover, this research seeks to put forward the relation between vocal cues, emotional regulation and empathic skills. 
Method
Sixteen forensic inpatients participated and signed an informed consent sheet. They were asked to recall five SDM, under recording, in accordance with the Self-Defining Memory Task (SDMT; Singer & Moffitt, 1992). Thereby, 87 SDM were collected. Then, each SDM was transcribed for the semantic and prosodic analyzes using respectively, EMOTAIX Text Analysis (Piolat & Bannour, 2009) and Praat (Boersma & Weeninck, 2013). Emotional and nonemotional sentences, defined using EMOTAIX classification, were compared over all SDM. Seven acoustic parameters were extracted: MeanF0, MedianF0, StandardDeviationF0, MinF0, MaxF0, Duration and SpeechRate. These measures were linked with the Cognitive Regulation Emotional Questionnaire (CERQ; Garfneski et al., 2001) and the Interpersonal Reactivity Index (IRI; Davis, 1980).

Results
Due to the absence of normality of distribution, we performed non-parametric comparison group analyzes. Our analyzes have shown that all antisocial participants expressed, in majority, neutral, specific, non-integrated SDM which related threating life event and interpersonal relations. AS participants exhibited more adaptative regulation strategies and less empathy than reference population standards. F0 variability was found significantly higher for emotional sentences, specific SDM, positive and mixed valence SDM. Moreover acoustic cues were correlated positively with adaptative regulation strategies and negatively with perspective-taking ability. These results will be discussed in light with literature.